Nishi Family › Compare › Surveys, Polls and Feedback
Nishi vs the Field
Surveys, Polls and Feedback — mechanically measured, liar-killed, sovereign.
Nishi Pulse vs SurveyMonkey and Google Forms and Typeform and Qualtrics (bars as of 2026-07)
How this is scored. Every Nishi cell is measured: the generator reads the real organ source on disk and requires the implementing symbol to exist (no self-grading). Competitor cells record documented capability presence. Best=leads this axis, Yes=present, Part=partial, No=absent. This is capability presence, not depth or scale: the majors lead on index size and neural ranking. Nishi's genuine exceeds are the sovereignty / neutrality / determinism axes.
Feature matrix
| Capability | Nishi | SurveyMonkey | Google Forms | Typeform | Qualtrics | Notes |
|---|---|---|---|---|---|---|
| Survey authoring as spec data | Yes | Yes | Yes | Yes | Yes | Rivals ship visual builders (deeper); ours is spec-as-data loaded over the live admin lane, never code |
| Question types: choice, scale, free text | Yes | Yes | Yes | Yes | Best | Three types today; Qualtrics question library (matrix, conjoint, heatmap) is the field's deepest |
| One-question quick poll | Yes | Yes | Yes | Yes | Yes | A poll is a 1-question survey on the same engine; LIVE at /survey/s/welfare-pulse |
| Anonymous-by-design responses | Yes | Part | Part | Yes | Yes | Spec @anon: tokens never rendered anywhere public (gate S6b); big rivals default to owner-visible respondent metadata |
| k-anonymity floor on public results | Best | No | No | No | Part | Aggregates withheld below @kmin quorum BY CONSTRUCTION, live-proven on the welfare survey; Qualtrics EX dashboards have anonymity thresholds |
| Revote updates, never double-counts | Yes | Part | Part | No | Part | Per-key latest-wins ballots (gate E5, live-proven n=3 stable through a revote); rivals offer opt-in response editing |
| Double-submit cannot inflate | Yes | Yes | Yes | Yes | Yes | Content-addressed records + latest-wins (gate E6) |
| Close and freeze results | Yes | Yes | Yes | Yes | Yes | Status re-stated as a new record version; history kept |
| Aggregate results with bars | Yes | Yes | Best | Yes | Yes | Counts, averages, distributions; Google Forms instant summary charts set the ease bar |
| Free text walled to admin report | Yes | Part | Part | Part | Part | Written answers NEVER on the public page by construction; rivals show open text to any result viewer |
| Admin lane fail-closed key gate | Yes | Yes | Yes | Yes | Yes | Absent key file = every admin call refused (gate S11); OPAQUE session auth = named rung |
| Personal respondent links | Yes | Yes | Part | Yes | Best | Links with ?t= give exact dedupe for employee and member audiences; Qualtrics XM directory deepest |
| Skip logic and branching | No | Yes | Yes | Yes | Best | Honest gap: flat questionnaires only today |
| Piping and carry-forward | No | Yes | No | Part | Best | Honest gap |
| Response quotas | Yes | Yes | No | No | Best | @quota N caps DISTINCT respondents; a revote by an existing respondent is EXEMPT (never inflates the tally), a NEW token past the cap is refused 409 and the open form withholds the questionnaire when full; gate 55/55 engine + 45/45 serve + LIVE-proven on nishifamily.com (fill-to-3 then 4th=409); Qualtrics per-segment/cross quotas deepest |
| Multi-language instruments | No | Yes | Part | Yes | Best | Honest gap; the CONNECT translation substrate is the named compose path |
| Email and channel distribution | No | Best | Yes | Yes | Yes | Honest gap: nx_send + vizsla wire = named rung; today links are shared by hand |
| Embeddable one-question poll | Yes | Yes | Yes | Best | Yes | Compact iframe poll GET/POST /survey/embed/<id> (cross-origin vote accepted, inline results bars, snippet on the results page); gate 38/38 + sovereign-verified; SurveyMonkey/Typeform embed-anywhere deeper |
| Offline capture | No | Part | No | No | Best | Honest gap |
| Results API JSON | Yes | Yes | Yes | Yes | Best | Versioned aggregate JSON at GET /survey/s/<id>/results.json, all-integer permille, k-floored, free text excluded (gate 34/34 + privacy tooth, sovereign-verified live); MCP tool = named rung |
| Welfare triage scoring and bands | Yes | Part | Part | Part | Yes | Composes the situation-management lib: data-driven weights band crisis vs stable (gate E12); rival scoring targets quizzes not welfare |
| Descriptive statistics: mean median stddev | Yes | Yes | Yes | Yes | Yes | Fixed-point integer, KAT-exact (stats gate 47/47); every rival computes these in dashboards |
| NPS with distribution and validity caveat | Yes | Yes | No | Yes | Best | Evidence-gated: banked 2024-26 studies show NPS validity actively studied, so the page forces distribution+mean beside it; Qualtrics CX suite deepest |
| Confidence-interval margins on public results | Best | No | No | No | Part | 95pct z-margin on the PUBLIC results page by default, small-n honest (live at /survey); Qualtrics computes CIs in its analysis tools not on shared summaries |
| Cross-tabulation with chi-square significance | Yes | Part | No | No | Best | Adversary-guarded: independent table = NOT significant, df past the critical table FAILS CLOSED (never false significance); live-proven chi 20.99 df 9 on the welfare survey |
| Key-driver correlation ranking | Yes | Yes | No | No | Best | Pearson vs the outcome question, ranked by strength (live: food +0.93 bills +0.90 vs spiritual support); adversary: uncorrelated data = exactly 0 |
| Segment breakdowns: means by group | Yes | Yes | Yes | Yes | Best | The improvement-area finder, live-proven: food security 8.5 under stable housing vs 1.0 in crisis |
| Trend across waves | Yes | Yes | No | Yes | Best | Before/after split on real ballot timestamps; banked repeated-cross-section studies are the evidence base |
| Open-text theme extraction | Yes | Yes | No | No | Best | Sovereign frequency themes (work, rent, childcare surfaced live from the welfare survey); LLM coding = the evidence-ranked next rung; Qualtrics Text iQ deepest |
| Open-text sentiment scoring (tone) | Yes | Yes | No | No | Best | VADER-style INTEGER compound (lexicon + negation-flip + intensifier + S/sqrt(S^2+a) norm): warming/cooling tone + positive/neutral/negative split per text question, admin-walled; gate 25/25 KAT-exact + adversary (negation flips 'not helpful' NEGATIVE, mixed nets neutral, gibberish 0) + evidence-gated (VADER needle in banked swc_svy_sent) + LIVE-proven on nishifamily.com; neural/LLM coding = the deeper named rung; Qualtrics Text iQ deepest |
| Triage rollup: crisis strained stable counts | Yes | No | No | No | Part | Welfare bands rolled up across all ballots (composes sv_triage); nearest rival analog is Qualtrics EX alerting |
| Attention-check (instructed-response) items | Yes | Yes | No | Yes | Best | Survey @check meta names a known-answer item; a failer is flagged careless (the C/IER gold standard, evidence-gated); Qualtrics ships attention checks + ExpertReview |
| Careless-response screening (C/IER) | Yes | Yes | No | No | Best | Attention-check + straight-lining detection, FLAG-not-drop; genuine low-but-varied answers protected (quality gate 14/14 + adversary); Qualtrics response-quality deepest |
| Evidence-gated metric selection | Best | No | No | No | No | Every shipped metric must find a verbatim needle in the banked 2024-26 literature or the row DIES (17/17, fabricated citation REJECTED); no rival grounds its metric roster mechanically |
| U1 Visual fidelity (bar: typeform.com 2026-07) | Yes | Yes | Yes | Best | Yes | Eyeball-honest: clean responsive cards + dark mode; Typeform's animated one-question flow is the field bar |
| U2 Interaction depth (bar: typeform.com 2026-07) | Yes | Yes | Yes | Best | Yes | Tappable radio scales; no drag-rank, autosave, or progress indicator yet |
| U3 Results information design (bar: surveymonkey.com 2026-07) | Yes | Best | Yes | Yes | Part | Bars, averages, distributions; SurveyMonkey crosstabs go deeper |
| U4 Product polish and feel (bar: typeform.com 2026-07) | No | Yes | Yes | Best | Yes | Honest zero: no onboarding, transitions, or themes -- the false-SOTA kill working as designed |
| Sovereign self-host zero per-seat | Best | No | No | No | No | Own language, compiler, TLS, store on owned hardware; none of the four rivals self-hosts at all |
| Additive never-lose ballot history | Best | No | No | No | No | Every ballot version preserved on the append-only store (gate E5e); a revote destroys nothing |
| Stable browser token, progressive enhancement | Yes | Yes | Yes | Yes | Yes | No-JS still works (fresh token per view, stated boundary); JS keeps one stable token per browser |
coverage 853/1000EXCEEDS 5PRESENT 30ABSENT 6
Honest verdict. Nishi has broad, sovereign search-engine presence — a real organ behind every core ranking and crawl axis, built bits-up in one language. It is not yet a scale or neural-ranking rival: the majors lead on index size (billions to trillions vs a Common-Crawl-seeded slice) and trained rerankers. Nishi's genuine exceeds are structural — owning the whole stack, neutrality by construction, and a zero-JS integer-deterministic result page. The climb is a bigger index and a trained reranker.
Generated by nx_swcompare_matrix (sovereign NishiLang organ) from knowledge/compare/surveys.matrix — every Nishi cell verified against real organ source on disk. Zero JS, zero trackers.